Automation inherits the structure of what came before it.

A conversation that has become familiar: a prospective client reaches out to talk about automating a part of their business. On the surface, the request is straightforward. They want a human-labor bottleneck relieved, they want consistency, they want throughput. They want AI, and they are ready to buy it.

The engagement diagnostic begins and, within a week, the real terrain emerges. There is no single workflow to automate. Different team members do the same work differently. Steps get skipped when the situation seems to call for it. Decisions are made on the fly, justified afterward, and rarely revisited. The documentation that exists describes a workflow nobody actually follows.

At which point a choice appears. We can build automation against the workflow as it is described, which will be confidently wrong. We can build against the workflow as it is actually practiced, which will hardcode every one of its informal compromises. Or we can do the thing the client did not ask for and help rebuild the workflow before anything is automated.

The reflex on the client side is resistance: they came for speed, not for a redesign. It is worth explaining why the redesign is not an overreach. Automation does not create structure; it inherits the structure of whatever came before it. If the existing workflow is held together by the judgment of a handful of senior operators, automation does not preserve that judgment — it amputates it and leaves the pieces running blindly.

The useful question, once we are inside the work, is rarely "what should we automate first?" It is closer to: what is the workflow we are about to bake into software for the next decade, and is it the workflow we want? Only after that question is answered does the automation conversation become serious.

We have come to believe that most of our job, in the first phase of any engagement, is not selecting a technology. It is helping a client see the work they already do, clearly enough that the system we build on top of it is built on something worth preserving.